The Quantis methodology — 8-layer deterministic pipeline
Quantis separates deterministic formulas, modeled benchmark ranges, scenario simulation, and AI narration. Its 97,500 benchmark cells are a rule-generated grid—not 97,500 observed companies—and dataset references are not linked one-to-one to individual cells.
Coverage and limitations
- Five tables of 19,500 modeled cells: core ratios, unit economics, geography, risk, and invoicing
- Country and region coverage is uneven; local conditions require direct evidence
- Grid cells have no per-row source, period, or quality-grade lineage
- Standard outputs do not constitute audited accounting, tax, investment, lending, or legal advice
The 8-Layer analysis pipeline
- Layer 1 — Input sanitization and business classification
- Layer 2 — Benchmark matching against a 97,500-cell modeled coverage grid; cells are not observed companies
- Layer 3 — Coarse location adjustment where supported
- Layer 4 — Deterministic financial calculations and quality checks
- Layer 5 — 1,000 scenario iterations with P10/P50/P90 modeled ranges
- Layer 6 — Classification, bounded narrative, action plan, and adversarial review
- Layer 7 — Reconciliation of supported numeric claims; dataset-level references exist, but per-cell source lineage does not
- Layer 8 — External evidence checks for selected metrics when suitable evidence is available
See the research · Security posture